Scalable Game Server Infrastructure

Role & context

As Lead & Senior Full Stack Engineer at RPA Nepal, I own the technical and DevOps decisions behind the game server infrastructure — designing it, keeping it running, and tuning it as load grows, alongside sprint planning and delivery for a team of 5 engineers.

The challenge

Game server workloads are spiky and latency-sensitive: player load shifts quickly, and a single overloaded instance directly translates into a bad player experience. The infrastructure needed to absorb 1,000+ concurrent users without manual intervention during load spikes, while keeping AWS spend under control rather than just over-provisioning for peak.

Approach

I designed the infrastructure around load balancing across server instances so no single node becomes a bottleneck under concurrent load, and paired that with an automated CI/CD pipeline — automated triggers plus one-click manual deploys — that cut deployment time from 3–4 hours down to 10 minutes. That deployment speed matters directly for infrastructure work: faster, lower-risk deploys mean load-balancing and capacity changes can ship and be verified quickly instead of being batched into rare, high-risk release windows.

I also built self-service DevOps tooling that lets new application instances be provisioned through a form rather than a manual ops request, reducing the operational overhead of scaling out the infrastructure as demand grows.

Tech stack

• AWS, with infrastructure cost tuning as an ongoing constraint, not an afterthought
• Load-balanced server architecture for concurrent player load
• Node.js backend
• Automated CI/CD (triggered + one-click manual deploys)

Results

• Infrastructure supports 1,000+ concurrent users via load balancing
• AWS costs optimized as part of the ongoing infrastructure design, not a one-off cleanup
• Deployment time cut from 3–4 hours to 10 minutes via CI/CD automation